“What predictive quality means to me is the ability to infer quality. If I know materials and components are within specs, and machines are operating within boundary conditions, then I know product quality will be OK.”

Abram Ziegelaar Head of Operations & Engineering Technology, B Braun

Increasing Product Recalls a Worrying Trend

A medical device recall is expensive, complex and disruptive. Class 1 devices pose the highest risk to patients’ lives where a failure or quality issue can potentially lead to death. From a business perspective, product recalls hurts sales and its reputation in the market. Considering the risk associated with medTech product recalls, manufacturers should be focused on quality assurance to prevent recall events. However, data from a recent post on medtechdive.com indicates the opposite. 

According to Sedgwick Brand Protection Recall Index, there were a total of 268 recalls in the second quarter, 34% higher than the first quarter. The total number of recalls were 10% higher than the quarterly average for the last five years, with 21 recalls being for Class 1 medical devices, with safety and software concerns being the top reasons. In the latest FDA guidelines on product recalls, medTech manufacturers need to be recall-ready with electronic means to ensure recalls are faster and more efficient. FDA also explains that voluntary recalls are the fastest, most effective way for a company to correct or remove potentially harmful products from the market.

The increasing number of product recalls clearly indicates that quality management in the medTech industry needs an overhaul. Predictive Quality is a promising new risk-based approach to quality management, which can potentially transform the system from being end-of-the-line-control based to all-pervasive-assurance based, incorporating quality in the very DNA of the product.

Predictive quality can prevent possible defects and can help predict recalls by combining AI, advanced analytics, design, production, supply-chain, and field data.

Understanding what changes with Predictive Quality

Predictive quality goes beyond traditional quality management approaches, by creating an end-to-end process envelope across the entire manufacturing operation. It is a risk-based approach that ensures products coming out in a batch/lot have been quality assured by meeting predetermined process conditions. When the process parameters remain within this performance envelope, it automatically releases products and reduces redundant testing and qualification efforts, automating quality management.

Figure 1: Source – Tech-Clarity

Traditional quality management systems are mostly paper or spreadsheets. Static tools at track events that have already occurred and ensures that compliance documentation is maintained in proper formats. Quality standard operating procedures are generally rigid and aimed at meeting compliance requirements and passing regulatory audits.

Predictive quality has created a paradigm shift in the way quality is perceived and implemented across a medtech value chain. The aim of the quality systems changes from reporting an incident to preventing it, delivering quality systems changes, and focuses on speed of decision making and profitability through quality improvement. Self-improving SOPs using AI and ML helps give operators assembly direction and has a direct and positive impact on product quality. It collates and contextualizes data from multiple sources and is analysed using the Manufacturing Data Platform to effect dynamic improvements to the SOPs.

Predictive quality allows medtech manufacturers improve their processes, personnel management, and material inputs so quality monitoring and manufacturing process are consistent. Quality assurance then becomes part of the process and not a separate exercise of quality control before shipping. Before implementing a quality management system, manufacturers need to understand what challenges need to be addressed and the role of the data platform for predictive quality to be successful.

The challenges in implementing Predictive Quality

Adapting to a risk-based approach- Implementing predictive quality hinges on process owners being able to adapt to a risk-based approach to quality management. A risk-based approach entails identifying and quantifying risks to patients and employees, then ascertaining how to minimize or eliminate those risks. Identifying critical to quality (CTQ) areas becomes challenging for companies with increasingly complex products, supply chains, and product variants. Creating and validating process flows is the best way to ensure success and extend predictive quality across the operation. A risk-based approach requires a data platform that enables the set-up of all possible process flows and activities.

Lack of feedback loops- Quality needs data coherence and for that to happen, accurate, complete and timely data must be provided for analysis and decision making. Regulators demand three levels of qualification; 1) installation, 2) operation, and 3) performance, (IQ, OQ, and PQ) to act as a feedback loop to ensure the process churns out acceptable products. As plants modernize and the volume of data increases, the data coming in from the plant and the field is put in context by the MES to make sense and gauge improvements for the future.

IT and OT convergence is a goal for many manufacturing enterprises but is hampered by data silos from multiple separate and disparate systems. Feedback loops are essential for predictive quality and medtech manufacturers can struggle establishing them. of the situation is exacerbated with complex supply chain partners, contract manufacturers, and multi-site manufacturing.

Information Disconnect- Both vertical and horizontal integration plays a critical role in predictive quality. Vertical integration can be defined as integrating an MES with the automation layer, the ERP, PLM, CRM, and SCM applications. the integration between MES applications at the plant level with each other, the extended supply chain partners, and the contract manufacturers represents horizontal integration.

Data from multiple sources creates an effective predictive quality model. This requires a data platform that provides integration across organizational layers including the supply chain, contract manufacturing, and customer network. When data from multiple sources are being contextualized through a single platform, the intelligence created is more pointed and helps create better predictive quality results.

Manufacturing Data Platform enables Predictive Quality

An integrated manufacturing data platform combines the plant-wide guidance and automatic documentation of MES with key quality capabilities, IIoT data capture, multi-stream data contextualization, and analysis.

Figure 2: Source – Critical Manufacturing

A data platform beyond the traditional MES, provides the means for a risk-based model of the entire manufacturing process, incorporating individual workflows and underlying activities, each categorized based on the risk associated with it. The platform provides IT and OT data convergence for the qualification process and ensures feedback loops are created and data enables effective and proactive quality management. The platform also vertically and horizontally integrates the value chain (see figure 2) and creates the basis of predictive quality deployment.

Medtech manufacturers can take the first step towards predictive quality with a scalable industry specific MES that can work as a full-function manufacturing data platform,. The right MES would provide the tools to bolster predictive quality, SPC, RCA, CAPA, and NCs modules. The data platform would bring integration and predictive analytics through AI to deliver the intelligence needed for proactive quality decisions that may save lives!